systematically passing around user rngs

[isaacsas]

In https://github.com/SciML/DiffEqJump.jl/pull/225 I couldn't consistently use the user-provided rng to JumpProblem as some places only have integrators as input (for example in callbacks). It would be nice to have a consistent interface for setting and getting at a rng across DiffEqJump (or even better SciML). Maybe stashing it in the integrator makes sense?